mirrorneuron-api 1.0.0

MirrorNeuron REST API

MirrorNeuron API

The RESTful HTTP Gateway for the MirrorNeuron distributed runtime system.

Architecture

Built with FastAPI and Uvicorn, this component acts as an HTTP abstraction over the core mn-python-sdk gRPC Client. It allows external microservices, web dashboards, or non-Python applications to easily interact with the MirrorNeuron cluster without speaking Protobuf.

Installation

Note: This API is installed automatically and symlinked globally as mn-api by the MirrorNeuron install.sh script.

pip install mirrorneuron-api

Running the Server

mn-api

This runs the Uvicorn server on port 4001 locally.

Configuration

All overrides use MN_ env vars:

• MN_ENV=prod requires MN_API_TOKEN.
• MN_API_HOST, MN_API_PORT control binding; default localhost:4001.
• MN_CORE_HOST controls the default core gRPC host; default localhost.
• MN_GRPC_TARGET or MN_CORE_GRPC_TARGET can override the full core gRPC target.
• MN_GRPC_TIMEOUT_SECONDS controls SDK calls.
• MN_API_REQUEST_SIZE_LIMIT_BYTES bounds request bodies.
• MN_API_CORS_ALLOW_ORIGINS is a comma-separated allowlist.

Protected endpoints accept Authorization: Bearer <MN_API_TOKEN>.

Endpoints

Method	Route	Description
GET	/api/v1/health	Service health check
GET	/api/v1/system/summary	Cluster hardware and pool state
GET	/api/v1/metrics	Runtime metrics summary
POST	/api/v1/jobs	Submit a new workflow via JSON
GET	/api/v1/jobs	List recent workflows
GET	/api/v1/jobs/{job_id}	Fetch workflow status
GET	/api/v1/jobs/{job_id}/events	Fetch job events
GET	/api/v1/jobs/{job_id}/dead-letters	Inspect dead-letter events
POST	/api/v1/jobs/{job_id}/cancel	Cancel a running job